Project Summary Understanding the coordinated activity of populations of neurons is a central goal in neuroscience. In no circuit are we closer to understanding this than in primary visual cortex (V1). However, despite decades of study our understanding of V1 is often restricted to responses to only special classes of stimuli, and we lack a theory that generalizes over stimuli to include the responses to natural images and movies. Our team proposal puts forth a program to dramatically broaden the scope of our understanding of V1 circuitry, with a deep focus on the vast array of inhibitory neurons. These data will drive a concerted modeling effort that engages in a virtuous back- and-forth with experimental projects where all projects share the goal of building a new circuit-based theory of visual processing. Internally generated variability in the cortex is a re?ection of its recurrent circuitry. However, most past mod- eling work has focused on impoverished circuits that collapse all sources of inhibition into stemming from one central pool of inhibitory neurons. This project will extend classic theories of recurrent cortical networks to include the rich diversity of inhibitory neurons that are found in mouse V1. This will include connectivity pro?les that cap- ture spatial and featured-based wiring, both between excitatory and diverse inhibitory neurons. Our new theory proposes to capture how internally generated population-wide shared variability depends upon circuit structure, and can be manipulated by the rich spatial and dynamic stimulus statistics associated with natural inputs. In particular, we will discuss the effective dimensionality of variability, something that contemporary theories are at a loss to explain. The high resolution calcium imaging data and optogenetic perturbations that are the focus of our experimental projects offer a unique opportunity to test and expand on our theories as they emerge. Finally, our modeling efforts will focus on how ?uctuations shape the transfer properties of neurons. This will serve as a platform to ground the ?ring rate models in our companion theory project. In sum, our project puts forth an ambitious program to build a theory of neuronal variability ? this is a critical ?rst step in building a circuit-based theory of visual processing in rodent V1.